Elektron-Users

Bencodec wrote:
i think the key to this trick is you'll need to get an LFO modulating the cutoff at a very fast rate, and then play with sweeping the cutoff filter manually or setup an another much slower lfo to also modulate it. crank the resonance up a bit and you should be in vowel sounding territory. not sure if the MD's lfos are fast enough, i do this on my voyage all the time with osc 3 as a modulator. i'll try this on the MD in a bit.

You're probably thinking of audio rate modulation, which the MD can't do, at least not with annoying artifacts. Audio rate modulation gives the impresson of two resonant peaks.

You'll get a cleaner and more pronounced result with two separate filters, which the MD can do quite easily.

I just thought, has anyone calculated the frequency values of the cutoff control of the MD's filters? I know the EQ frequencies have been recorded, but has someone done the same with the filters? I'm currently doing a long practical investigation in A2 Physics which is based on analysing and resynthesizing (via additive resynthesis) human speech, or at least formants. I could transfer some of these frequencies over to the MD if I had the correct values to tune into...

futureimage wrote:
I just thought, has anyone calculated the frequency values of the cutoff control of the MD's filters? I know the EQ frequencies have been recorded, but has someone done the same with the filters? I'm currently doing a long practical investigation in A2 Physics which is based on analysing and resynthesizing (via additive resynthesis) human speech, or at least formants. I could transfer some of these frequencies over to the MD if I had the correct values to tune into...

Not that I know of.

Using the technique described arlier, with the same signal passing through several INP machines with the filter set to bandpass, it's fully possible to do a "manual" formant synthesis of up to 16-parts. All the 16 parts would have identical settings apart from the cutoff frequency. You'd have to manually program the volume envelope for each part, which admittedly would be a pain. You'd have tremendous control, though. The MD filter is very well suited for this kind of processing, since it's quite neutral.

Set up identically the MD can also make up the synthesis section of a 16-band vocoder. Not sure how you would perform the analysis though. The only input-dependent processing the MD can do that I can think of now is the envelope follower in one of the input machines. This applies to the filter cutoff, not the volume.

It's a shame you can't route the individual MD parts in series internally, like on the MnM.

Nils wrote:
Set up identically the MD can also make up the synthesis section of a 16-band vocoder.



I might try some of these techniques out with the MnM, the melodic synth engines could be more musically useful, though I did do some multi-tracking with the MD doing this stuff and create a pretty cool avant-garde composition earlier.

sounds like Predator!!

Here's an abstract composition I recorded pretty quickly using these techniques. Enjoy

Sounds amazing..

I am really curious how this is been done..
Is it rude to ask for a a sysex file?

I'm afraid I can't do SysEx right now, my set up won't allow me (I can never get the Sysex to transfer correctly and I think I removed the settings for this, I'll just reset them when I need to).

Following this should get you somewhere though:

Three band pass filters are used:
-one on the source sound (this is very quiet in the mix as there was little of a vowel effect achieved)
-one on Input A (panned hard left)
-one on Input B (panned hard right)

The routing is as follows:
-Source -> (BP Filter) -> Input A -> (BP Filter) -> Input B -> (BP Filter)
-There is a little feedback from Input B's filter back into Input A for extra resonance

LFOs control the center frequency, bandwidth and resonance of each filter.